A practical approach to the management of cerebral amyloid angiopathy

Abstract

Cerebral amyloid angiopathy (CAA) is a common small vessel disease in the elderly involving vascular amyloid-β deposition. CAA is one of the leading causes of intracerebral hemorrhage (ICH) and a significant contributor to age-related cognitive decline. The awareness of a diagnosis of CAA is important in clinical practice as it impacts decisions to use lifelong anticoagulation or nonpharmacological alternatives such as left atrial appendage closure in patients who have concurrent atrial fibrillation, another common condition in older adults. This review summarizes the latest literature regarding the management of patients with sporadic CAA, including diagnostic criteria, imaging biomarkers for CAA severity, and management strategies to decrease ICH risk. In a minority of patients the presence of CAA triggers an autoimmune inflammatory reaction, referred to as CAA-related inflammation (CAA-RI), which is often responsive to immunosuppressive treatment in the acute phase. Diagnosis and management of CAA-RI will be presented separately. While there are currently no effective therapeutics available to cure or halt the progression of CAA, we discuss emerging avenues for potential future interventions.

Introduction

Cerebral amyloid angiopathy (CAA) is characterized by amyloid-β deposits in the walls of leptomeningeal and cortical blood vessels^1,2. CAA can lead to symptomatic lobar intracerebral hemorrhages (ICH) as well as smaller regions of bleeding including cerebral microbleeds (CMBs) and cortical superficial siderosis (cSS). There are a few genetic mutations that result in a pure form of CAA, characterized by early occurrence of ICH and in some cases cognitive dysfunction¹. However, in the great majority of patients, CAA is a sporadic disease that shows some overlap with Alzheimer’s Disease (AD) pathology.

In addition to being a cause of lobar ICH, sporadic CAA is an important etiology of cognitive impairment and gait instability in the elderly³. While CAA is classically thought of as a disease leading to cerebral bleeding, recent work has demonstrated that CAA also induces widespread ischemic changes including cortical microinfarcts and white matter hyperintensities, pathologies which are associated with cognitive decline^4,5.

There are currently no effective treatments available to cure or halt the progression of CAA. Management of patients mostly centers around reducing the risk of first-time or recurrent lobar ICH. This review provides an up to date and comprehensive discussion of MRI markers of CAA, particularly delving into the utilization of these markers to determine a patient’s individual risk of future ICH. Strategies for ICH risk reduction in CAA will be discussed in depth, including the management of frequently comorbid cardiovascular risk factors, with a particular focus on the avoidance of systemic anticoagulation and potential alternatives to anticoagulation.

Additionally, this review includes a discussion of emerging potential future therapies for CAA, ranging from decreasing production of amyloid-β to enhancing amyloid-β clearance.

Diagnosis of CAA

While direct visualization of vascular amyloid-β in CAA requires brain biopsy or autopsy, a set of clinical and MRI-based diagnostic criteria termed the (modified) Boston Criteria have been validated to enable the diagnosis of CAA without the need to obtain brain tissue^6,7. These criteria indicate that the presence of cortical hemorrhagic lesions including lobar ICH, strictly cortical CMBs, and/or cSS without the presence of deeper hemorrhages in patients >55 years of age is highly sensitive and specific for CAA. CMBs and cSS can be observed with gradient echo and susceptibility-weighted MRI sequences as shown in Figure 1. CMBs are found more frequently in posterior brain regions but they can be seen anywhere in the cortex in the presence of CAA⁸. Recent work has also demonstrated that the presence of strictly superficial cerebellar microbleeds is associated with CAA^9,10.

Figure 1. Hemorrhagic and non-hemorrhagic lesions associated with CAA.

Representative scans from multiple patients with probable CAA are shown. Top row (left to right): Non-contrast CT scan, susceptibility-weighted MRI, susceptibility-weighted MRI, susceptibility-weighted MRI. Bottom row (left to right): T2-weighted FLAIR MRI, T2-weighted MRI, diffusion-weighted MRI, T2-weighted FLAIR MRI. Arrows point to representative examples of each lesion in each panel.

A number of non-hemorrhagic imaging findings are also suggestive of a diagnosis of CAA and scale with CAA severity including white matter hyperintensities (WMH), MRI-visible perivascular spaces (PVS), lobar lacunes, and cortical microinfarcts⁴. An updated version of the Boston Criteria (version 2.0), which takes into account these non-hemorrhagic findings is expected to be released soon¹¹. White matter hyperintensities (WMH) are lesions visible on T2-weighted fluid-attenuated inversion recovery (FLAIR) MRI (see Figure 1), strongly associated with small vessel disease and thought to be in part ischemic in etiology. When severe, WMH can also be seen on CT as hypoattenuating lesions. These lesions are observed either around the ventricles or more widespread across the white matter centrum semiovale. WMH in the centrum semiovale and posterior regions of the brain are particularly associated with CAA^12,13. The severity of WMH correlates with vascular amyloid-β burden¹⁴, number of CMBs¹², lobar ICH risk^15,16, plasma amyloid-β 40 levels¹⁷, and cognitive impairment (as discussed further below). The extent of these lesions is classically assessed using the Fazekas grading scale¹⁸, though computer assisted segmentation techniques are also utilized for research studies^17,19. The presence of more than 10 subcortical WMH spots in one hemisphere (not affected by ICH) was also found to be significantly more common in patients with CAA when compared to hypertensive arteriopathy, a feature that can suggest CAA diagnosis in the appropriate setting⁸.

MRI-visible PVS are interstitial-fluid filled spaces surrounding blood vessels which appear as small elongated structures on T1- or T2-weighted MRI sequences²⁰ (Figure 1). These spaces can be differentiated from prior areas of infarction by the absence of a surrounding T2 hyperintense ring on FLAIR²⁰. A recent study comparing patients with CAA-related ICH to those with hypertensive arteriopathy-related ICH found that MRI-visible PVS located in the centrum semiovale were significantly more prevalent in patients with CAA-related ICH, while MRI-visible PVS at the level of the basal ganglia were indicative of hypertensive arteriopathy. The degree of centrum semiovale MRI-visible PVS was associated with known hemorrhagic imaging markers of CAA including cSS and CMBs²¹.

Along the same lines, it was recently observed that lacunar infarcts in the centrum semiovale (and not the basal ganglia), so-called ‘lobar lacunes’, are more commonly found in patients with CAA compared to hypertensive arteriopathy^22,23 (Figure 1). The presence of lobar lacunes correlates with increased risk of recurrent ICH in CAA patients²⁴. Although the pathophysiological mechanisms remain to be elucidated, this imaging marker may be a useful marker in the management of CAA.

Microinfarcts are small ischemic lesions, a subset of which can be observed with 7T or even conventional (i.e. 1.5–3T) MRI in the cortical grey matter (Figure 1)²⁵. Neuropathological studies have shown, however, that MRI drastically underestimates the number of cortical microinfarcts^26–28. In CAA, microinfarcts are associated with regional CAA severity and may indicate more severe vascular amyloid-β deposition²⁶. Cortical microinfarcts are thought to be important contributors to cognitive dysfunction in CAA and other cerebral microangiopathies.

Finally, diagnostic tools to assess for amyloid-β can also be supportive of a CAA diagnosis, particularly in cases with diagnostic uncertainty between CAA and/or hypertensive arteriopathy as the etiology of hemorrhage²⁹. Differentiating between CAA and hypertensive arteriopathy is crucial as recurrent ICH risk differs significantly³⁰. Hypertensive arteriopathy typically leads to hemorrhage in deeper locations as opposed to CAA. Some patients present with mixed location ICH and CMBs (deep and lobar). Recent studies have suggested that hypertensive arteriopathy is the predominant pathology in these mixed ICH/CMB cases²⁹.

The two most commonly used techniques to assess amyloid-β levels in the brain are cerebrospinal fluid (CSF) samples and amyloid-PET scans. The levels of both amyloid-β₄₀ and amyloid-β₄₂ isoforms are decreased in the CSF of patients with CAA or Alzheimer’s disease (AD), whereas tau levels are increased³¹. CAA is associated with predominantly vascular deposition of amyloid-β₄₀ isoforms, while AD is associated with parenchymal deposition of predominantly amyloid-β₄₂. Although CSF analysis alone does not sufficiently differentiate between CAA and AD, studies do suggest significantly lower levels of CSF amyloid-β₄₀ in patients with CAA as compared to those with AD^31,32. Importantly, CSF analysis in individuals with a hereditary form of CAA (i.e. Dutch-type CAA) reveals that decreased amyloid-β₄₀ levels in CSF is an early feature in the disease pathogenesis³³. Amyloid-PET tracers including Florbetapir and ¹¹C-Pittsburgh compound (B) (PiB) label both parenchymal and vascular amyloid-β. Amyloid PET may be useful as an adjunct test for CAA diagnosis in cognitively healthy people and to differentiate CAA from hypertensive arteriopathy particularly in patients with only microbleeds (no ICH) (see Figure 2), a pattern of non-hemorrhagic MRI markers strongly suggestive of CAA, or mixed location ICH/microbleeds^34–37. In fact, a recent study demonstrated that, in cognitively normal patients, florbetapir-PET can be used to diagnose probable CAA with high sensitivity (100%) and specificity (90%)³⁴. However, like CSF markers of amyloid-β, differentiation between vascular and parenchymal amyloid-β deposition and thus between CAA and parenchymal AD pathologies is challenging.

Figure 2. Florbetapir-PET scans from representative patients with CAA and hypertensive arteriopathy.

Left: Scan from a patient with CAA demonstrating decreased gray-white contrast, indicating increased gray matter florbetapir uptake (positive scan). Right: Scan from a patient with hypertensive arteriopathy demonstrating clear gray-white contrast throughout (negative scan). HTN = hypertension. [Adapted from Gurol et al., Neurology 2016]

Intracerebral hemorrhage risk-stratification and risk-reduction

Lobar ICH is a significant cause of morbidity and mortality in patients with CAA³⁸, and much of the management of CAA centers around reducing a patient’s risk of future ICH. Understanding a patient’s individual ICH risk is essential for both the counseling of patients and families and determining the risk-benefit balance for anticoagulation agents.

1. Risk-stratification

1.1. Cortical microbleeds and ICH risk

A large portion of patients with CAA present for neurological care with mild cognitive impairment, gait instability, or transient neurological symptoms and are subsequently diagnosed with CAA through an MRI of the brain demonstrating cortical CMBs but no larger hemorrhages. Making a diagnosis of CAA with these isolated imaging findings is crucial - these patients are at significantly higher risk for ICH than the general population^30,39. Additionally, establishing a diagnosis at this stage may prevent the need for further extensive work-up and may enable patients to participate in future trials at an early timepoint in their disease course.

In a recent observational study, patients diagnosed with CAA with CMBs (with no history of ICH) were found to have a 5% annual risk of ICH³⁹. This risk is significantly higher than the incidence of ICH in the general population which is ~0.025%⁴⁰ (and higher than the annual risk of recurrent ICH in patients with hypertensive arteriopathy-associated ICH which is ~ 2%⁴¹. Additionally, a study of 4,759 participants with cortical and/or deep CMBs followed over an 8 year period found that the risk of hemorrhagic and ischemic stroke increased with number of CMBs⁴². Patients in this study with cortical CMBs without deep CMBs (consistent with a diagnosis of CAA) were at a higher risk of ICH than those with solely deep CMBs.

For patients with CMBs and a prior lobar ICH, annual risk of recurrent lobar ICH increases to approximately 10%³⁰. As discussed above, non-hemorrhagic imaging features such as the extent of WMH, MRI-visible PVS in the centrum semiovale, lobar lacunes, and cortical microinfarcts are suggestive of more severe CAA and are associated with increased ICH risk^15,16,43.

1.2. Cortical superficial siderosis and ICH risk

cSS has recently emerged as one of the strongest predictors of future ICH^44,45. cSS refers to the presence of blood products in the subarachnoid space and superficial cortical layers, represented on MRI by curvilinear susceptibility artifact lining the sulci. These findings are thought to represent leakage of blood products from leptomeningeal blood vessels into the subarachnoid space and/or the chronic manifestation of prior sulcal subarachnoid hemorrhage. cSS often presents clinically as transient focal neurological episodes (TFNEs), which are discussed in detail below.

A patient’s risk of ICH scales with the multifocality and extent of cSS; a recent prospective study of 313 patients with CAA-related ICH found that annual risk of ICH increases with the extent of cSS, with annual ICH recurrence rates of 26.9% in patients with bilateral disseminated cSS⁴⁴.

The mechanisms connecting cSS and ICH remain unclear. Pathological studies observed that cSS is associated with increased local leptomeningeal amyloid-β but with decreased global cortical amyloid-β^46,47. In contrast, CMBs are associated with increased global cortical vascular amyloid-β²⁶. These findings, in addition to associations between cSS and apolipoprotein E allele ∈2 (as discussed further below), have led to the hypothesis that there may be different phenotypes of CAA including 1) patients with more prominent leptomeningeal disease and cSS (and potentially higher ICH risk) and 2) patients with more prominent cortical CAA and CMBs^26,48. Of note, the Piedmont mutation form of hereditary form of CAA resulting from a mutation in amyloid precursor protein (Leu705Val) is associated with prominent leptomeningeal CAA with relative sparing of the cortical vessels and typically presents with a severe lobar ICH phenotype^49,50. These findings support the notion that severe leptomeningeal CAA puts patients at risk for both cSS and ICH.

1.3. Apolipoprotein E allele status and ICH risk

Apolipoprotein E alleles ∈2 (APOE2) and ∈4 (APOE4) have both been associated with CAA and risk of lobar ICH^51–55. APOE4 is associated with increased parenchymal and vascular amyloid-β and is the strongest genetic risk factor for late onset AD, while APOE2 may track more closely with the severity of vascular pathology and ICH risk^48,56. A recent meta-analysis demonstrated an association between cSS severity and APOE2⁵⁷ and prior neuropathological work also identified a relationship between cSS and APOE2⁴⁶. However, these relationships have yet to be fully elucidated.

2. Risk-reduction

Reducing the risk of ICH in CAA predominantly centers around managing comorbidities including cardiovascular risk factors and decision-making surrounding anticoagulation (see Figure 3).

Figure 3. Managing comorbidities in patients with CAA to reduce intracerebral hemorrhage risk.

Recommendations for management of cardiovascular risk factors in patients with CAA (A); Possible alternatives to lifelong anticoagulation in patients with CAA, based on the clinical indication (B). Individualized decisions regarding anticoagulation should be made through multidisciplinary discussions, involving patients and families. Key: BP = blood pressure; ACC/AHA = American College of Cardiology/American Heart Association; DVT = deep vein thrombosis; PE = pulmonary embolism; IVC = inferior vena cava; LAAC = left atrial appendage closure; APLAS = antiphospholipid antibody syndrome.

2.1. Long-term blood pressure management

Blood pressure management is critical for reducing the risk of CAA-related ICH as well as hypertensive angiopathy-related ICH⁵⁸. An observational study of patients with lobar and deep ICH found an association between elevated blood pressures and recurrent ICH risk in both groups⁵⁹. While the severity of hypertension correlated with ICH recurrence risk, the study also demonstrated an association between blood pressures in the pre-hypertensive range (systolic 120–139 mmHg/ diastolic 80–89 mmHg) and ICH risk. These findings and work from others suggest patients with a diagnosis of CAA should ideally have blood pressures maintained at <120/80 mmHg⁶⁰.

In addition to pre-hypertension being a risk factor for ICH, an long-term trajectory of increasingly elevated blood pressures may contribute to ischemic stroke and ICH risk⁶⁰. Visit-to-visit variability in blood pressures has also been shown to be associated with CMB and WMH progression⁶¹ as well as all-cause mortality, cardiovascular disease, and stroke⁶².

2.2. Lipid management

Aggressive lipid management is an effective form of secondary ischemic stroke prevention⁶³. However, reduced low-density lipoprotein (LDL) levels (less than 70–80 mg/dL) have been associated with an increased risk of ICH^64–66. ICH patients randomized to high dose statin had a significantly elevated risk of recurrent ICH in the SPARCL study⁶³. A recent prospective study followed 96,043 participants with no prior history of stroke or myocardial infarction over 9 years of follow-up and found that participants with LDL levels <70 mg/dL had a significantly higher risk of ICH (both lobar and deep ICH)⁶⁷. A recent analysis of patients from the longitudinal, prospective Framingham study demonstrated an association between statin use and deep ICH, but not with lobar ICH⁵⁴. Interestingly, a recent observational study of 345,531 patients followed for 9.5 years demonstrated a decrease in ICH risk with statin use, while still finding an association between lower LDL levels and increased ICH risk⁶⁸.

Given the potential benefits of statins in cerebral small vessel disease, current recommendations are to pursue statin therapy in patients with CAA who have a clear indication per ACC/AHA guidelines. However, benefits and risks of statin use should be discussed with all patients at a higher ICH risk^66,69. The SATURN study, an NIH-funded randomized controlled trial (RCT), will assess the potential benefit/harm of continuation versus discontinuation of statins in patients who survived a lobar ICH. This study will hopefully provide a much-needed answer to the dilemma of statin use in CAA⁷⁰.

2.3. Glucose management

A recent prospective study of 96,110 participants demonstrated that both low (<4 mmol/L) and high (>7 mmol/L) fasting blood glucose levels are associated with increased ICH risk; however, the association between high fasting blood glucose and ICH risk was only found for patients with deep ICH⁷¹. The number of lobar ICHs in this study was limited. Additionally, a subgroup analysis of the INTERACT2 study demonstrated a strong association between hyperglycemia and poor outcomes post-ICH, although most patients in this study had a deep location of ICH⁷². More work is needed to further delineate the relationship between blood glucose levels and ICH risk in CAA.

2.4. Antiplatelet use

Current guidance is that antiplatelet agents for primary prevention of cardiovascular events should be avoided in patients with CAA unless the patient has a clear, evidence-based indication for antiplatelet use⁷³. Prior studies have demonstrated increased CMBs^74,75 and ICH rates^16,76 with antiplatelet use.

However, antiplatelet agents do offer significant benefits for cardiovascular disease and should be used in patients with CAA in situations in which there is a clear indication such as secondary prevention or other FDA-approved situations such as after coronary or cardiac procedures. The RESTART study was a prospective randomized trial of 537 patients who presented with an ICH while on antiplatelet therapy for secondary prevention⁷⁷. Patients were randomized to either restart or avoid antiplatelet therapy. Investigators did not find a significant difference in ICH recurrence between the two groups, leading them to conclude that the benefits of antiplatelet therapy for secondary prevention likely exceed the ICH-related risks.

2.5. Anticoagulation

Anticoagulation significantly increases the risk of ICH and ICH-related mortality, even in patients with low baseline ICH risk. The risk of ICH increases by 2–5x with anticoagulation with warfarin and non-vitamin K oral anticoagulants (NOACs)^78,79. While hemorrhagic complications may occur less frequently with NOACs than with warfarin^80–83, this finding may not apply to patients at high risk for ICH as these patients were excluded from all trials comparing NOACs to warfarin. The 3-month mortality of patients with anticoagulation-associated ICH is approximately 50% for patients taking warfarin^84,85 or NOACs^80,81.

Because of the risk of ICH, ideally lifelong anticoagulation should be avoided in all patients with CAA; however, there are cases in which the potential benefits of anticoagulation may outweigh the risks, especially for short-duration anticoagulant use. We note that there have been no randomized controlled trials to date of anticoagulation use in patients with known CAA. Therefore, for all patients at risk for both ischemic and hemorrhagic events, patients and families should be engaged in shared-decision making discussions regarding potential initiation of anticoagulation for all indications discussed below.

Decision-making regarding initiation or continuation of anticoagulation in patients who are at both high ischemic and hemorrhagic risk depends on both an individual patient’s risk of ICH, the indication for anticoagulation, potential alternatives to anticoagulation, and the required duration of anticoagulation as outlined below⁷⁸. A recent study analyzed data from 2 longitudinal cohort studies of patients with anticoagulation-associated ICH utilized MRI markers (cSS and CMBs) and APOE genotype to create a risk stratification tool for recurrent ICH⁵⁵, which may aid in clinical decision-making.

2.5.1. Non-valvular atrial fibrillation

One of the most common indications for anticoagulation is for ischemic stroke prevention in patients with atrial fibrillation⁸⁶. Clinically, the CHA₂DS₂-VASC score is typically used to assess the annual risk of ischemic stroke in a patient with atrial fibrillation and to determine whether anticoagulation is indicated⁸⁷. Some providers use the HAS-BLED score to assess the risk of major bleeding events secondary to anticoagulation, however, this score has poor predictive value overall and it was not designed to be used in patients with significant ICH risk. A recent multicenter observational study showed that HAS-BLED score’s predictive value for ICH was less than 50%, (i.e. worse than the flip of a coin), so it is not used in the field of neurology⁸⁸. In general, anticoagulation, especially long-term anticoagulant use, should be avoided in all patients with CAA, if possible. As mentioned above, patients and families should be engaged in shared-decision making discussions regarding potential initiation of anticoagulation in patients at risk for both ischemic and hemorrhagic events.

Left atrial appendage closure (LAAC) has recently emerged as a viable alternative to anticoagulation, and should be considered in patients with elevated ICH risk⁷⁸. In patients with atrial fibrillation, over 90% of thrombi form in the left atrial appendage and therefore LAAC may help prevent thrombus formation and subsequent ischemic stroke⁸⁹. Multiple devices have been developed for LAAC, and the currently only FDA-approved device with the largest amount of data supporting its use is the WATCHMAN device (Boston Scientific, Marlborough, MA). LAAC via the WATCHMAN has been shown to be non-inferior to warfarin and apixaban at ischemic stroke prevention^90–92. Other devices include 1) the AMPLATZER AMULET (St Jude Medical/Abbot, Minneapolis, MN) device, for which the IDE RCT comparing it to WATCHMAN was completed and results are currently pending, 2) the AtriClip which can be used for LAAC during open heart surgery (Atricure Inc, Westchester, Ohio)⁹³, and 3) the LARIAT suture delivery system (SentreHeart, Redwood City, CA). CHAMPION AF (the new generation WATCHMAN FLEX) and CATALYST (AMPLATZER AMULET) are ongoing RCTs comparing the outcomes of LAAC to NOACs in the general nonvalvular atrial fibrillation populations. LAAC and ICH risk stratification strategies were also reviewed in detail⁷⁸.

Initial studies of the WATCHMAN device excluded patients with past history of ICH as these studies had a long-term warfarin arm and a short course of warfarin was used for LAAC (typically ~45 days for the WATCHMAN) to allow for device endothelialization to prevent thrombi formation on the device surface⁹⁴. However, studies have demonstrated non-inferiority with NOACs for short-term anticoagulation post LAAC^94,95, and dual antiplatelet therapy can be used as an alternative to short-term anticoagulation post-procedurally^96,97. It is important to note the difference between short term (~6 weeks post LAAC) anticoagulation vs lifelong anticoagulation as the risk of ICH increases over years. Decisions regarding the use of short-term anticoagulation or dual antiplatelet therapy should be made in conjunction with a cardiology team based on a risk versus benefit analysis accounting for a patient’s individual ICH risk (as discussed above).

2.5.2. Mechanical valves

Patients with mechanical valves require life-long anticoagulation, and warfarin continues to be the only approved anticoagulation agent for this indication. Given the significant risk of thrombus formation on mechanical valves, the benefits of anticoagulation in this scenario outweigh the risks of ICH even in patients with CAA. However, in patients with particularly high-risk as discussed above (e.g. multifocal cSS with history of prior ICH), depending on the clinical scenario, one may need to consider replacement with a bioprosthetic valve.

2.5.3. Intracardiac thrombus, pulmonary embolism, deep vein thrombosis

The presence of a new/fresh intracardiac thrombus is an absolute indication for anticoagulation given the associated risk of ischemic stroke. However, in patients with CAA, one can consider short-term repeat imaging to assess for thrombus resolution in order to minimize time spent on anticoagulation.

In cases of pulmonary embolism (PE), treatment depends on the clinical severity of the PE balanced with a patient’s individualized risk of ICH. Short-term anticoagulation is allowed in most situations, but decision-making should be multi-disciplinary.

Lastly, in patients diagnosed with a deep vein thrombosis (DVT), inferior vena cava (IVC) filters can be considered as a short-term alternative to anticoagulation in patients at high risk for ICH. A short-term course of anticoagulation with interval repeat imaging to assess for DVT resolution can also aid the decision-making process.

2.5.4. Antiphospholipid antibody syndrome

Patients with antiphospholipid antibody syndrome (APLAS) have a significant risk of arterial and venous thrombosis and require life-long anticoagulation with warfarin⁹⁸. Unfortunately, there is no alternative to anticoagulation for this patient population. Despite the ICH risk, anticoagulation should be continued as long as the diagnosis of APLAS is appropriately established.

Management of transient focal neurological episodes

Patients with CAA frequently experience TFNEs, intermittent episodes of transient neurological symptoms (<24 hours), including sensory symptoms (paresthesias and numbness), focal weakness, and language disturbances⁹⁹. TFNEs can be stereotyped and are sometimes described as having a spreading progression. While the pathophysiology remains largely unknown, these are thought to represent cortical spreading depolarizations¹⁰⁰, potentially triggered by regions of cSS and also associated with acute convexity subarachnoid hemorrhage^99,101. There is limited data available for treatment of TFNEs in CAA. Providers often trial anti-epileptic medications which have been shown to be effective at CSD reduction in migraineurs^102,103 such as valproic acid, lamotrigine, and topiramate in patients with recurrent TFNEs with anecdotal benefit.

Of note, TFNEs can be misdiagnosed as transient ischemic attacks (TIAs) and one should consider obtaining gradient echo or susceptibility-weighted MRI to assess for CMBs and cSS in patients with focal neurological deficits. Antithrombotics should not be started or escalated if the correct diagnosis is TFNEs.

Management of cognitive impairment

Many patients with CAA present with mild cognitive impairment. Cortical microinfarcts have been associated with cognitive impairment independent of AD pathology both in CAA and other disease processes^5,25,28,104. Neuropathological studies have demonstrated that microinfarcts occur in brain regions with increased amyloid-β burden and are associated with vessels with significant amyloid-β deposition¹⁰⁵. In addition to microinfarcts, white matter hyperintensities and atrophy and structural network alterations have been shown to contribute to cognitive impairment in patients with CAA^106–108.

There are currently no effective treatments for these cognitive symptoms in CAA patients. Given these symptoms may represent a relatively early manifestation of CAA (preceding lobar ICH); prompt diagnosis, potentially through early detection of microinfarcts, may provide a window for early intervention.

Management of CAA-related inflammation

A subset of patients with CAA develop episodes of spontaneous inflammation. The most common presenting symptoms of CAA-related inflammation (CAA-RI) are headache, seizures, focal neurological deficits, and subacute to acute cognitive decline^109–111. MRI features typically include 1) asymmetric WMH which extend subcortically, 2) hemorrhagic lesions including CMB and cSS and/or 3) post-contrast leptomeningeal enhancement^109,110,112. CSF profiles are generally inflammatory, with a lymphocytic pleocytosis and elevated protein, but these findings are not always present^109,113. Interestingly, during acute episodes of CAA-RI anti-amyloid-β autoantibodies are found in the CSF, suggesting a spontaneous immune-mediated response to vascular amyloid-β in the brain¹¹⁴. These imaging and clinical findings resemble amyloid-related imaging abnormalities (ARIA) in the context of anti-amyloid immunotherapy trials¹¹⁵.

CAA-RI involves predominantly perivascular inflammation, whereas a related condition amyloid-β related angiitis (ABRA) involves both perivascular and transmural inflammation and has many similarities to CNS vasculitis^116–118. ABRA is an angiodestructive process and, in addition to the above imaging features, will commonly present with acute microinfarcts.

Patients with CAA-RI typically present with lower rates of lobar ICH than patients with “non-inflammatory” CAA at initial diagnosis¹¹². However, long-term strategies to reduce risk of future ICH (as discussed above) should be utilized in these patients as well.

The treatment of CAA-RI relies on intensive immunosuppression, predominantly with steroids¹¹⁹. A typical treatment course involves a high-dose intravenous steroid pulse followed by a prolonged steroid taper (at least six months). Some providers use cyclophosphamide in addition to steroids, particularly in cases of ABRA given its similarities to CNS vasculitis^116,119,120. Both CAA-RI and ABRA respond well to initial treatment in over 75% of patients^{111,119–121}. In fact, if there is no clinical and radiographic improvement to an initial steroid pulse, an alternate diagnosis should be considered.

Relapse rates remain largely unknown given the relative rarity of the disease and inconsistent treatment protocols. A recent retrospective cohort study of 48 individuals with CAA-RI observed a recurrence rate of 26% in patients treated with immunosuppressive agents as compared to 71% in patients not treated with immunosuppression¹¹⁹. Relapses can occur at the initial site of inflammation and/or in additional sites¹²². In the setting of relapse (or in patients having difficulty tolerating long-term steroids), steroid-sparing agents such as mycophenolate and azathioprine can be considered.

Potential emerging treatments for CAA

There is a great unmet need for the development of novel therapeutic strategies aimed at halting or slowing down the progression of disease in CAA patients. In the final section of this review, we briefly discuss recent advancements in the development of novel candidates targeting the pathogenic cascade of events initiated by amyloid-β. These approaches mainly revolve around amyloid-lowering strategies at different stages of disease pathogenesis, including (antibody-mediated) removal of aggregated or soluble forms of amyloid-β, reducing amyloid-β production, and nonpharmacological strategies to enhance amyloid-β clearance through perivascular drainage pathways (see Figure 4).

Figure 4. CAA pathogenesis and potential targets for intervention.

Amyloid-β (Aβ) deposits and precursors shown in blue; potential interventions at each stage shown in orange. A tangential cross-section of a diving arteriole is depicted; pial surface to the right of the figure.

Antibody-mediated removal of amyloid-β from the brain has long been the focus of attention in the AD field. Given the overlap in disease pathophysiology, a similar approach has been considered for patients with CAA¹²³. Lessons learned from immunotherapy trials in AD patients, however, have important implications for the translation to patients with CAA particularly with respect to safety considerations. Initial studies to remove amyloid-β from the brains of patients with AD focused on active immunization¹²⁴. While this early attempt did demonstrate effective removal of amyloid-β plaques, 18/298 (6%) of treated patients developed meningoencephalitis, which prompted trial discontinuation^125,126. These findings were consistent with ARIA¹²⁷ and have been linked with the presence of CAA on neuropathological examination¹²⁸. Since then, the focus has shifted to passive immunotherapy with anti-amyloid antibodies, yielding disappointing results. A notable exception are recent reports from the phase 3 Aducanumab trial suggesting slowing of cognitive decline in AD patients treated with Aducanumab compared to placebo controls^129–131. Yet, as in previous passive immunotherapy trials, ARIA were commonly observed, in as many as 47% of patients receiving the highest dose^129–131. Indeed, ARIA occur more frequently in patients with APOE4 and cortical CMBs on MRI, suggesting that co-existing CAA increases the risk of ARIA in AD patients¹²⁷. Insights from neuropathological investigations suggest that removal of amyloid-β from severely affected vessels in the context of advanced CAA may predispose these vessels to bleeding, which is in line with observations in patients with CAA-RI²⁶. Therefore, immunotherapy may not be the safest candidate approach for patients with CAA. To date one anti-amyloid immunotherapy trial has been performed in patients with CAA, which yielded negative results¹³². Interestingly, although patients did not develop ARIA, vascular reactivity (the primary outcome marker) was found to be reduced in patients who received the antibody compared to placebo controls, suggesting worsening of vascular function in the short-term. Since patients were only followed for ninety days, it remains currently unknown what the potential long-term effects are of immunotherapy in patients with advanced CAA and whether immunotherapy directed at early disease stages might prove to be more beneficial (and safer).

Therapies can also be targeted at soluble amyloid-β removal. Tramiprosate is a low molecular weight agent shown to bind to soluble amyloid-β and effectively decrease vascular amyloid-β in mouse models¹³³. This compound was tested in the first reported clinical trial of a candidate treatment in CAA patients, and demonstrated safety and target engagement¹³⁴. It remains unclear whether this approach can effectively halt disease progression and prevent future ICH in patients with CAA. Increasing proteolytic degradation of amyloid-β is another potential therapeutic option, with the enzyme neprilysin demonstrating potential efficacy in mouse models¹³⁵.

An alternative approach is targeting amyloid-β production to halt the cascade of events leading up to vascular amyloid-β accumulation. Such a prevention strategy may be particularly effective in pre-symptomatic mutation carriers of hereditary forms of CAA. Inhibitors of the β-site amyloid cleaving enzyme (BACE) have long been the leading option, but unfortunately recent trials in AD patients were halted due to cognitive worsening. Work is currently underway to develop anti-sense oligonucleotides against the amyloid precursor protein (APP), awaiting future results¹³⁶.

Lastly, diminished clearance of amyloid-β through perivascular drainage pathways is thought to play an important role in the pathophysiology of CAA¹²³. Low-frequency spontaneous oscillations of the arterial vasculature, also known as vasomotion, are a potential driving force for clearance of waste products from the brain, including amyloid-β¹³⁷. These low frequency arteriolar oscillations are particularly apparent during slow-wave sleep and have been shown to be coupled to CSF flow dynamics in humans¹³⁸. Interestingly, enhancing vasomotion through sensory-evoked vascular reactivity was recently shown to increase clearance of fluorescent tracers from the mouse brain¹³⁷. Further research is needed to assess whether vasomotion can be enhanced in patients with CAA, either through non-invasive sensory stimulation or promoting healthy sleep, and whether this may improve clearance of amyloid-β.

Conclusions

CAA is a small vessel disease which leads to lobar ICH, cognitive impairment, and TFNEs in older adults. Increasing numbers of studies have provided biomarkers to diagnose CAA and predict the risk of first-time and recurrent lobar ICH. The availability of non-anticoagulant stroke prevention strategies has empowered physicians to circumvent the need for lifelong anticoagulant use even in patients with nonvalvular atrial fibrillation. Despite these promising advances in ischemic stroke and ICH prevention, there are currently no effective disease-modifying treatments to slow down or halt the progression of CAA. More work is urgently needed to develop novel therapeutic strategies for CAA.